Intratumor lactate levels reflect HER2 addiction status in HER2‐positive breast cancer

Despite different molecular tumor profiles indicate that human epidermal growth factor receptor 2 (HER2) messenger RNA (mRNA) levels mirror HER2 addiction and trastuzumab benefit in HER2‐positive breast cancer (BC), the identification of noninvasive clinical predictors of trastuzumab sensitivity remains an unmet clinical need. In the current study, we investigated whether intratumor lactate levels reflect HER2 addiction and, in turn, trastuzumab susceptibility. Accordingly, the gene expression profiles of transgenic murine BC cell lines expressing the human d16HER2 variant (HER2‐addicted) or human full‐length HER2 (WTHER2; HER2‐nonaddicted) revealed a significant enrichment of glycolysis‐related gene pathways in HER2‐addicted cells. We studied the metabolic content of 22 human HER2‐positive BC by quantitative nuclear magnetic resonance spectroscopy and found that those cases with higher lactate levels were characterized by higher HER2 transcript levels. Moreover, gene expression analyses of HER2‐positive BC samples from a TCGA data set revealed a significant enrichment in glycolysis‐related pathways in high/HER2‐addicted tumors. These data were confirmed by metabolic analyses of human HER2‐positive BC cell lines with high or low HER2 transcript levels, which revealed significantly more active glycolytic metabolism in high HER2 transcript than in low HER2 transcript cells. Overall, our results provide evidence for noninvasive intratumor lactate detection as a potential metabolic biomarker of HER2 addiction and trastuzumab response suggesting the possibility to use in vivo imaging to assess lactate levels and, in turn, select HER2‐positive BC patients who are more likely to benefit from anti‐HER2 treatments.

possibility to use in vivo imaging to assess lactate levels and, in turn, select HER2-positive BC patients who are more likely to benefit from anti-HER2 treatments.  Loibl & Gianni, 2017;Ménard et al., 2000).
Globally, HER2-positive BC accounts for approximately 20% of all BCs and is characterized by a highly aggressive disease course, more frequent relapses in the absence of effective adjuvant treatments, and poor survival in the metastatic setting (Loibl & Gianni, 2017;Ménard et al., 2000). The introduction of the humanized anti-HER2 monoclonal antibody (mAb) trastuzumab in combination with chemotherapy has improved both the relapse-free survival (RFS) and overall survival (OS) of primary and metastatic HER2-positive BC patients (Joensuu et al., 2006;Piccart-Gebhart et al., 2005;Pupa et al., 2005;Romond et al., 2005;Slamon et al., 2001). In recent years, different rationally designed HER2-targeting agents have been developed and successfully tested in clinical trials. Among them, the combination of trastuzumab with lapatinib (a tyrosine kinase inhibitor) or pertuzumab (anti-HER2 mAb) improved the pathological complete response rate in the neoadjuvant setting compared with trastuzumab alone (Baselga et al., 2012;Gianni et al., 2012), while the Aphinity trial demonstrated a reduced risk of relapse in patients who received the trastuzumab-pertuzumab combination in the adjuvant setting (Von Minckwitz et al., 2017).
Unfortunately, the response of HER2-positive tumors to anti-HER2 treatments is highly heterogeneous, with some patients deriving long-term benefit and other patients bearing primarily resistant tumors (Carey, 2012). High HER2 expression, as determined by IHC score or FISH analysis, cannot be considered per se a good indicator of HER2 addiction or the response to trastuzumab or other anti-HER2 treatments because of (a) poor performance of IHC and FISH methods currently used to define HER2-positive BC, which do not take into account HER2 protein levels or provide suboptimal quantification of HER2 levels, and (b) molecular heterogeneity of HER2-positive neoplasms, with some tumors relying on other growth-proliferation pathways either independent or downstream of HER2 signaling. Distinguishing between HER2-addicted and HER2nonaddicted BC, that is those responsive or not responsive to trastuzumab, could help clinicians to select HER2-positive BC patients who are more likely to benefit from anti-HER2 therapies in the adjuvant and metastatic settings.
Recent studies have suggested that high HER2 messenger RNA (mRNA) levels are directly associated with the trastuzumab response, and thus, high HER2 mRNA levels have been proposed as a marker of HER2 addiction (Di Modica et al., 2017;Prat et al., 2014;Triulzi et al., 2015). Several preclinical studies have revealed the capability of HER2-driven signaling to promote glucose uptake, oxygen consumption, overexpression of lactate dehydrogenase-A (LDH-A), and lactate production in tumor cells (Walsh et al., 2013;Zhao et al., 2009). In particular, intratumor lactate accumulation has been associated with tumor metastatic potential and patient OS and has been shown to contribute to immune escape, cell migration, and radioresistance in several studies (Hirschhaeuser et al., 2011). In addition, HER2-induced upregulation of hypoxiainduced factor 1α (HIF1α; Laughner et al., 2001) could mediate the burst in glycolytic metabolism that is a feature typical of these tumors (Laughner et al., 2001;Robey et al., 2005) (Thermo Fisher Scientific, Waltham, MA;last verification, November 2015). The transgenic murine primary mammary tumor cell lines MI6, MI7, WTHER2_1, and WTHER2_2, which express the human d16HER2 splice variant or full-length wild-type HER2 (WTHER2), have been described previously (Castagnoli et al., 2014;Castagnoli et al., 2017;De Giovanni et al., 2014); these cell lines were maintained in complete MammoCult medium (StemCell Technologies, Vancouver, BC, Canada) supplemented with 1% FBS (Sigma-Aldrich) and penicillin-streptomycin (Sigma-Aldrich). All tumor cell lines were cultured at 37°C in a humidified 5% CO 2 atmosphere and were routinely tested for mycoplasma contamination.

| Patient cohort
The 39 HER2-positive BC patients included in this study were part of the observational retrospective multicenter Italian study Group Herceptin in Adjuvant Therapy (GHEA; Campiglio et al., 2013;Supporting Information

| Western blot analysis
Human HER2-positive BC cell lines were solubilized for 40 min at 0°C with lysis buffer, as described (Ghedini et al., 2010). Briefly, after solubilization, the samples were mixed with gel sample buffer under reducing conditions, heated for 5 min at 95°C, and resolved by electrophoresis on precast 4-12% Bis-Tris gels (Thermo Fisher Scientific).
The separated proteins were electrophoretically transferred onto nitrocellulose membranes, which were stained with Ponceau S to check protein loading, washed extensively with Tris-buffered saline (TBS)+ 0.5% Tween-20 and incubated for 1 hr at room temperature in blocking solution (5% low-fat milk in TBS + 0.1% Tween-20) before the addition of primary antibodies in 3% low-fat milk in TBS + 0.1% Tween-20 for 1 hr at room temperature or overnight at 4°C with gentle shaking. The following primary mouse mAbs were used: Ab3 c-erbB-2/HER2/neu (1:100; Calbiochem, Darmstadt, Germany) directed to the human HER2 intracellular domain and anti-β-actin-peroxidase (AC-15 clone; 1:50,000; Sigma). The membranes were then washed extensively with TBS + 0.5% Tween-20 and incubated with horseradish peroxidase-conjugated goat anti-mouse IgG (1:5,000; Amersham GE Healthcare, Little Chalfont, UK) for 1 hr at room temperature. The signals were detected using enhanced chemiluminescence (ECL; Amersham GE Healthcare).

| Assessment of extracellular lactate levels
Extracellular lactate levels in conditioned medium from cultured cancer cells were determined by a blood gas analyzer (GEM Premier 4000; Werfen, Bedford, MA). Briefly, HER2-positive BC cell lines were cultured in a monolayer until they reached approximately 80% confluence; 12 hr before the test, the culture medium was replaced with fresh medium. To avoid protein contamination that could alter the lactate quantification, culture media were deproteinized using Amicon Ultra centrifugal filter units (molecular weight cut-off 10 kDa; Merck-Millipore, Billerica, MA) following the manufacturer's instructions.
Lactate levels were normalized to the number of tumor cells.

| Bioinformatic analyses
Raw gene expression data for murine mammary cell lines and human tumor specimens (GHEA cohort) were downloaded from the Gene Expression Omnibus (GEO) repository with accession numbers GSE67300 and GSE55348, respectively. Data from both datasets were log 2-transformed and normalized using the robust spline normalization method implemented in the Bioconductor package lumi (Du et al., 2008).
Only probes with a detection p < 0.01 in at least one sample were retained, and for each gene, the probe with the highest variance was selected. Samples in GSE55348 were assigned to PAM50 subtypes ( (Ritchie et al., 2015). Gene set enrichment analysis (GSEA) (Subramanian et al., 2005) was performed in a pre-ranked mode using gene sets from the MSigDB C2 canonical pathway collection. For GSE67300, genes were ranked according to the t-statistic calculated between the d16HER2 and WTHER2 cell lines using the limma package [24]. For TCGA, genes were ranked according to the Pearson's correlation coefficient with ERBB2 gene expression levels.

| High-resolution MRS of the cell and tissue extracts
Aqueous extracts from 20 to 30 × 10 6 cells grown to 60-70% confluence and tissues were prepared in EtOH:H 2 O (70:30, vol/vol) as previously described (Iorio et al., 2010). Samples were ultrasonicated at 20 kHz with an exponential probe (8 mm  High-resolution MRS analyses (25°C) were performed at 9.4 T (Bruker Avance spectrometer, Karlsruhe, Germany). Spectra of the cell and tissue extracts were obtained using radio-frequency pulses for excitation, water signal presaturation, data processing, and data analysis as described (Pisanu et al., 2014). Relative metabolite quantification was presented in nanomole normalized to the number of extracted cells or as the percentage of an individual metabolite among all metabolites.

| Statistical analysis
Associations between categorical variables were tested using the twosided Fisher's exact test. A p < 0.05 indicated statistical significance.
Differences in continuous variables between groups were tested using a two-tailed unpaired t test. Differences were considered significant at Enhanced glycolytic metabolism and upregulation of the expression of glycolytic enzymes in human cancer is frequently sustained and stimulated by overexpression of HIF1α (Hockel & Vaupel, 2001;Marin-Hernandez et al., 2009;Semenza, 2003). In this context, it is reported that HIF1α expression is regulated by the HER2 signaling (Laughner et al., 2001). We, therefore, used qPCR analysis to assess a potential correlation between HER2 transcript or lactate levels and HIF1α expression (Hockel & Vaupel, 2001;Semenza, 2003). In a cohort of 39 frozen HER2-positive BC specimens that included 18 previously tested cases, we found a positive correlation between the transcript levels of HER2 and HIF1α (p = 0.0092 and Pearson's r = 0.4171; Figure 3a). As shown in Figure 3b Furthermore, we observed a significant direct correlation between HER2 mRNA and NADH levels (Figure 5d), between HER2 mRNA levels and the optical redox ratio (Figure 5e). These findings indicate that HER2 overexpression is associated with the upregulation of glycolytic metabolism. Then, we analyzed the global metabolic content of HER2-positive BC cell lines by high-resolution MRS; in accordance with previous results, we detected a significant enrichment of intracellular lactate levels in BC cell lines with high HER2 mRNA expression (Figure 6a). We also found a direct correlation between HER2 transcript levels and lactate content (Figure 6b). Since glycolytic cancer cells produce and release lactate in the extracellular microenvironment, we measured lactate levels in conditioned media from matched HER2-positive BC cell lines (Supporting Information Figure   3). This analysis revealed significantly higher lactate levels in conditioned media from BC cell lines with higher HER2 mRNA expression. In contrast, intracellular levels of succinate (Supporting Information Figure 4), a metabolite produced in the TCA cycle and associated with active oxidative phosphorylation, were significantly higher in HER2-positive BC cell lines characterized by lower HER2 mRNA levels, and HER2 mRNA expression was inversely associated with intracellular succinate content. Taken together, our data support the hypothesis that high HER2 levels are associated with the upregulation of glycolysis compared with oxidative mitochondrial phosphorylation, resulting in lactate production and excretion by cancer cells.

| DISCUSSION
In this study, we provide the first evidence that intratumor lactate levels and glycolytic metabolism reflect addiction of HER2-positive BC models to HER2 signaling, and are associated with higher HER2 mRNA levels. We confirmed these findings through gene expression and metabolic analyses of HER2-positive cancer cell lines, transgenic models, and human tumor specimens.
Although the introduction of anti-HER2 therapies has dramatically improved the cure rate and survival expectancy of patients with early stage and advanced-stage HER2-positive BC, respectively, tumor sensitivity to anti-HER2 treatments is highly heterogeneous (Loibl & Gianni, 2017). In recent years, specific gene expression signatures have been found to be associated with the tumor response to trastuzumab (Castagnoli et al., 2014;Prat et al., 2014;Triulzi et al., 2015). HER2 mRNA levels have also been proposed to reflect HER2 expression and to predict trastuzumab efficacy better than the currently used IHC and FISH techniques (Pogue-Geile et al., 2013;Prat et al., 2014;Triulzi et al., 2015). Our in silico data strongly support this hypothesis and confirm that HER2 mRNA levels represent an easy-to-measure, low-cost and semiquantitative biomarker that reflects the HER2 addiction status and trastuzumab susceptibility. However, HER2 mRNA quantification will be difficult to implement in standard clinical practice due to concerns with the quality of tumor specimen fixation, embedding and deparaffinization, which are crucial steps for the reliable quantitative analysis of HER2 mRNA levels.
Previous studies have reported enhanced glycolytic metabolism in HER2-positive BC and HER2-mediated expression of glycolysisrelated genes (Walsh et al., 2013). In particular, HER2 signaling is known to upregulate LDH-A and HIF1α expression (Zhao et al., 2009), while HER2 downregulation or pharmacologic inhibition with trastuzumab reverses the glucose dependency and inhibits glycolytic metabolism in HER2-positive BC cells (Walsh et al., 2013).
However, to the best of our knowledge, the upregulation of glycolysis has not been correlated with tumor response to anti-HER2 agents. The major finding of our study is the association between enhanced glycolytic metabolism and addiction to HER2 signaling in HER2-positive BC. We also found an association between the

| 1777
The potential role of lactate quantification in predicting tumor sensitivity to anticancer treatment and patient survival has been previously investigated. For example, Blatt et al. (2016) reported that lactate levels in head and neck squamous cell carcinoma specimens inversely correlated with patient OS and PFS after surgery and radiation therapy. The measurement of intratumor lactate levels with noninvasive methodologies, such as nuclear magnetic resonance, was previously proposed as a method to identify key biomolecules and molecular changes associated with cancer aggressiveness and treatment and to improve clinical cancer care (Haris et al., 2015). Our study adds to this knowledge by establishing a link between intratumor lactate levels and addiction to HER2 signaling in HER2-positive BC that predicts the benefit from anti-HER2 treatments.
The strengths of our study include the following: the identification of a metabolic signature associated with addiction to HER2 signaling, thus paving the way for noninvasive intratumor lactate measurements to identify patients who will benefit from anti-HER2 therapies; and (b) the consistency of our results in preclinical models (both in vitro and in vivo) and in tumor specimens. The major limitation of the study is the small number of tumor specimens analyzed in the validation patient cohort.
The identification of lactate quantification as a novel, noninvasive, and reliable metabolic biomarker of addiction to HER2 signaling may improve the selection of HER2-positive BC patients who are more likely to benefit from standard anti-HER2 treatments, while unselected patients are steered towards enrollment in clinical trials.
Larger studies are required to confirm that intratumor lactate is a Ghirelli for technical assistance, the personnel of the Tissue Bank of the Fondazione IRCCS Istituto Nazionale dei Tumori, Milan, for providing BC frozen samples, and Laura Mameli for secretarial assistance.

CONFLICTS OF INTERESTS
All authors declare that there are no conflicts of interest.